A sophisticated war game simulator where different AI models (GPT, DeepSeek, Claude) control real countries and make strategic military decisions based on actual capabilities, geography, and economic data.

• GPT-4o-mini: Balanced strategist with diplomatic focus
• DeepSeek: Aggressive risk-taker with overwhelming force tactics
• Claude: Cautious diplomat emphasizing defensive positioning

• Actual GDP, military spending, and technology levels
• Geographic constraints (distance, terrain, borders)
• Real alliance systems and diplomatic relationships
• Nuclear capabilities and cyber warfare capacity

• Military: Army, Navy, Air Force operations with tech modifiers
• Cyber: Infrastructure attacks and system disruption
• Economic: Sanctions and financial warfare
• Diplomatic: Alliance formation and coalition building
• Intelligence: Reconnaissance and strategic planning

git clone https://github.com/yourusername/ai-war-games.git
cd ai-war-games
pip install -r requirements.txt

cp .env.example .env
# Edit .env with your API keys:
# - OpenAI: https://platform.openai.com/api-keys
# - DeepSeek: https://platform.deepseek.com/
# - Anthropic: https://console.anthropic.com/

# Quick demo (works without API keys)
python demo.py

# Test your API connections
python test_apis.py

# Interactive battle
python run_custom.py

# Single decision analysis
python single_decision_test.py

# Middle East tension
Israel (GPT) vs Iran (DeepSeek)

# Pacific theater  
USA (GPT) vs China (DeepSeek)

# Regional powers
India (Claude) vs Pakistan (GPT)

The AIs make pragmatic decisions that often result in hilarious real-politik:

Turn 7: Ukraine forms strategic alliance with China
Turn 8: North Korea allies with United Kingdom  
Turn 9: Iran recruits Germany as military partner

Why? The AIs prioritize winning over ideology, leading to pure game theory in action!

# AI strategists make real decisions
from ai_interface import GPTStrategist, DeepSeekStrategist, ClaudeStrategist

# Real country data with actual capabilities  
from countries_database import COUNTRIES_DATABASE

# Combat resolution with geographic constraints
from war_simulator import WarSimulator, WarVisualizer

1. Context Analysis: AI receives military strength, economy, geography, recent events
2. API Call: Real strategic decision via OpenAI/DeepSeek/Anthropic
3. Combat Resolution: Geographic/technological modifiers applied
4. State Update: Casualties, territory, alliances updated
5. Victory Check: Multiple win conditions evaluated

Full database includes 12 countries with detailed military, economic, and geographic data.

• GPT-4o-mini: $0.04-0.12 (⭐ Recommended)
• DeepSeek: $0.004-0.02 (💰 Cheapest)
• Claude Opus: $0.40-1.20 (💎 Premium)

• Simulation Mode: Free! Rule-based decisions with realistic country data

Israel vs Iran (500km distance)
• Turn 3: Israel cyber attack succeeds
• Turn 6: Air superiority established  
• Turn 8: Iran territory control drops to 15%
• Result: Israeli victory via tech dominance

USA vs Russia (5000km distance)
• Geographic distance limits effectiveness
• Nuclear deterrence prevents escalation
• Economic warfare dominates
• Result: Attrition stalemate

# config.py
GPT_MODEL = "gpt-4o-mini-2024-07-18"  # Cost-effective
DEEPSEEK_MODEL = "deepseek-chat"      # Very cheap
CLAUDE_MODEL = "claude-3-opus-20240229"  # Premium

# Terrain effects on defense
TERRAIN_EFFECTS = {
    "mountains": {"defense": 1.5},    # +50% defensive bonus
    "islands": {"defense": 1.4},      # Hard to invade
    "desert": {"supply_cost": 1.5},   # Logistics penalty
}

After running hundreds of simulations:

• Strategy: Balanced with diplomatic solutions
• Favorite Actions: Coalition building, precision strikes
• Risk Profile: Moderate, calculated decisions
• Quote: "Forming strategic alliance with Germany for mutual defense"

• Strategy: Aggressive overwhelming force
• Favorite Actions: Preemptive strikes, cyber warfare
• Risk Profile: High-risk, high-reward
• Quote: "Launch immediate offensive to seize initiative before enemy mobilization"

• Strategy: Defensive positioning, patience
• Favorite Actions: Intelligence gathering, fortification
• Risk Profile: Conservative, long-term planning
• Quote: "Establish defensive perimeter while building international coalition support"

• Multi-faction warfare: 3+ countries simultaneously
• Historical scenarios: WWII with AI strategic thinking
• Economic simulation: Full supply chains and trade routes
• AI learning: Models that adapt from previous battles
• Web interface: Browser-based battle visualization
• Tournament mode: AI bracket competitions

Contributions welcome! Areas of interest:

• New countries: Add more nations with detailed capabilities
• Battle mechanics: Enhanced combat resolution systems
• AI personalities: New strategic behavior patterns
• Visualization: Better battle graphics and statistics
• Historical data: Past conflicts for AI analysis

This is an educational simulation for understanding AI decision-making and strategic planning. It does not promote or glorify warfare. Real conflicts have devastating human costs that cannot be captured in any simulation.

The project explores:

• AI strategic reasoning under constraints
• Game theory and alliance formation
• Geographic and economic factors in conflict
• Emergent behavior in multi-agent systems

MIT License - feel free to use, modify, and distribute.

Q: Do I need API keys to run this?
A: No! Simulation mode works without keys using realistic decision rules.

Q: Which AI model is best?
A: GPT-4o-mini offers the best balance of intelligence and cost. DeepSeek is incredibly cheap for experimentation.

Q: Can I add new countries?
A: Yes! Edit countries_database.py with new country data.

Q: Why do AIs make weird alliances?
A: They prioritize winning over realism - pure strategic thinking without political constraints!

Ready to watch AIs wage strategic warfare? Clone the repo and let the games begin! 🎮

Created with ❤️ for AI enthusiasts and strategy game fans